Steps toward a neural theory of self-actualization.
    The qualitative psychological notion of self-actualization, due to Maslow, is understood mechanistically as an optimal state that is available to, but not always reached by, a neural network.  This is described in terms of a Lyapunov (energy) function for a submodule of the network, combined with a supervising node that sends a signal if the energy function for the submodule's current state is higher than the energy function for other imagined states.  Suggestions are made for parametric variations that determine to what extent the network will change its state if the current state is found not to be globally optimal, even if it is optimal for a smaller subnetwork.  Speculative analogies are drawn between parts of the network and functions of the amygdala, prefrontal cortex, and midbrain noradrenergic system.